Personal portfolio and blog of Haowei Chan, featuring projects, writing, and web experiments.https://haoweichan.github.io/https://haoweichan.github.io/blog/when-every-node-succeeds/https://haoweichan.github.io/blog/when-every-node-succeeds/A reader said "every post has a fair few typos", but that month no node in the pipeline had failed once. Six defects, none of them a failure: every one was a component doing exactly what it was told, returning 200, then going wrong in the inch it was never told about. Plus the question I now keep on my review list: is this success signal load-bearing?Wed, 02 Sep 2026 00:00:00 GMThttps://haoweichan.github.io/zh-tw/blog/when-every-node-succeeds/https://haoweichan.github.io/zh-tw/blog/when-every-node-succeeds/讀者說「每一篇都有不少錯字」，但那個月 pipeline 沒有任何一個節點失敗過。六個缺陷，沒有一個是失敗：全都是某個元件正確完成了它被交代的工作、回傳 200，然後在它沒被交代的那一寸出錯。以及我因此加進 review 清單的那個問題：這個成功訊號，是承重的嗎？Wed, 02 Sep 2026 00:00:00 GMThttps://haoweichan.github.io/blog/factor-research-internal-controls/https://haoweichan.github.io/blog/factor-research-internal-controls/Researcher degrees of freedom are the real adversary in quantitative research. From predeclaration, orthogonality prescreens, and Gate A to the neutralization ladder and a forward shadow book: a control process I have run for over a year and violated a few times myself. It does not guarantee profit; it guarantees I can tell when a result is fake.Thu, 06 Aug 2026 00:00:00 GMThttps://haoweichan.github.io/zh-tw/blog/factor-research-internal-controls/https://haoweichan.github.io/zh-tw/blog/factor-research-internal-controls/研究者自由度是量化研究最大的敵人。從預先聲明、正交前篩、閘門 A、中性化階梯到前瞻影子簿：一套實際跑了一年多、也實際被自己違反過幾次的內控流程。它不保證賺錢，只保證結果是假的時候看得出來。Thu, 06 Aug 2026 00:00:00 GMThttps://haoweichan.github.io/blog/media-secrets-cloud-exit/https://haoweichan.github.io/blog/media-secrets-cloud-exit/Large immutable files and small highly sensitive config are the two poles a database migration never covers: an in-place storage-layer swap, a three-tier secret resolution order, half-finished-work archaeology, and the discipline of export, off-site backup, reconcile, only then delete.Wed, 05 Aug 2026 00:00:00 GMThttps://haoweichan.github.io/zh-tw/blog/media-secrets-cloud-exit/https://haoweichan.github.io/zh-tw/blog/media-secrets-cloud-exit/大體積不可變檔案與高敏感小體積設定，是資料庫遷移沒覆蓋到的兩極：存儲層原地替換、secrets 三層解析順序、半成品考古學，以及「export → 異地備份 → 核對 → 才刪」的刪除紀律。Wed, 05 Aug 2026 00:00:00 GMThttps://haoweichan.github.io/blog/document-db-to-postgres-migration/https://haoweichan.github.io/blog/document-db-to-postgres-migration/How to cut a migration that touches every read and write path into six independently shippable, independently reversible phases: same-interface swaps, the dual-write safety net, write-stop as the only irreversible point, and the fourteen production bombs adversarial review caught.Tue, 04 Aug 2026 00:00:00 GMThttps://haoweichan.github.io/zh-tw/blog/document-db-to-postgres-migration/https://haoweichan.github.io/zh-tw/blog/document-db-to-postgres-migration/把一場「動所有讀寫路徑」的遷移，切成六個可以獨立上線、獨立回滾的階段：同介面替換、雙寫安全網、停寫作為唯一的不可逆點，以及對抗審查抓出的十四個生產級炸彈。Tue, 04 Aug 2026 00:00:00 GMThttps://haoweichan.github.io/blog/cloud-bill-forensics/https://haoweichan.github.io/blog/cloud-bill-forensics/The bill went from a few hundred NTD to over forty thousand, and nobody broke in — every cent was spent by my own code. A five-layer drill-down from invoice total to function name, and why payload x distance costs more than read count.Mon, 03 Aug 2026 00:00:00 GMThttps://haoweichan.github.io/zh-tw/blog/cloud-bill-forensics/https://haoweichan.github.io/zh-tw/blog/cloud-bill-forensics/帳單從幾百元變成四萬多，沒有人入侵——每一分錢都是自己的程式碼花掉的。從總金額一路下鑽到函式名的五層偵查法，以及「payload × 距離」為什麼比讀取次數更貴。Mon, 03 Aug 2026 00:00:00 GMThttps://haoweichan.github.io/blog/broker-reports-and-portfolio-automation/https://haoweichan.github.io/blog/broker-reports-and-portfolio-automation/Resident Personal Assistant Series: Parsing encrypted PDF statements, integrating multi-broker APIs, handling T+2 settlement lags, and building automated research report crawlers.Sun, 15 Mar 2026 00:00:00 GMThttps://haoweichan.github.io/zh-tw/blog/broker-reports-and-portfolio-automation/https://haoweichan.github.io/zh-tw/blog/broker-reports-and-portfolio-automation/個人常駐助理系列：解密加密 PDF 對帳單解析、多經紀商 API 串接、T+2 交割款時差計算與自動化券商研報爬蟲系統設計。Sun, 15 Mar 2026 00:00:00 GMThttps://haoweichan.github.io/blog/resident-personal-assistant/https://haoweichan.github.io/blog/resident-personal-assistant/Building a resident assistant on Hermes Agent: six traits that stop it being a chatbot, what is genuinely clever about the Hermes agent mechanism, and the day it became trustworthy.Tue, 03 Mar 2026 00:00:00 GMThttps://haoweichan.github.io/zh-tw/blog/resident-personal-assistant/https://haoweichan.github.io/zh-tw/blog/resident-personal-assistant/用 Hermes Agent 蓋一個常駐、會自己開口的個人助理：六個讓它不像聊天機器人的特性、Hermes 的 agent 機制聰明在哪，以及它變得可信的那一天。Tue, 03 Mar 2026 00:00:00 GMThttps://haoweichan.github.io/blog/fastapi-redis-tiered-cache-design/https://haoweichan.github.io/blog/fastapi-redis-tiered-cache-design/Full-Stack Architecture Series: Not all APIs deserve the same TTL. How to combine Edge CDNs, Redis in-memory caching, and Cache-Aside patterns to prevent Cache Stampedes.Fri, 20 Feb 2026 00:00:00 GMThttps://haoweichan.github.io/zh-tw/blog/fastapi-redis-tiered-cache-design/https://haoweichan.github.io/zh-tw/blog/fastapi-redis-tiered-cache-design/全端產品架構系列：並非所有 API 都適用相同的 TTL。如何結合 Edge CDN、Redis 記憶體快取與 Cache-Aside 模式應對 Cache Stampede 並提升系統吞吐量。Fri, 20 Feb 2026 00:00:00 GMThttps://haoweichan.github.io/blog/dev-staging-prod-environments/https://haoweichan.github.io/blog/dev-staging-prod-environments/Environment layering is not enterprise ceremony, but a risk-reduction pattern: how to boost release reliability using branch-to-environment mapping, preview deploys, secrets isolation, and production guardrails.Tue, 10 Feb 2026 00:00:00 GMThttps://haoweichan.github.io/zh-tw/blog/dev-staging-prod-environments/https://haoweichan.github.io/zh-tw/blog/dev-staging-prod-environments/環境分層不是大型公司的儀式，而是一個降低發布風險的設計模式：如何透過 Branch-to-env 映射、Preview Deploys、Secrets 隔離與生產環境 Release Guardrails 提升發布可靠性。Tue, 10 Feb 2026 00:00:00 GMThttps://haoweichan.github.io/blog/agentic-ingestion-pipeline-part2/https://haoweichan.github.io/blog/agentic-ingestion-pipeline-part2/Production lessons: Step-by-step checkpoints and rerun_from recovery, preventing reasoning model token truncation, pre-persistence gates, and concurrency budgets.Thu, 22 Jan 2026 00:00:00 GMThttps://haoweichan.github.io/zh-tw/blog/agentic-ingestion-pipeline-part2/https://haoweichan.github.io/zh-tw/blog/agentic-ingestion-pipeline-part2/生產環境實戰：Step-by-step Checkpoints 重播機制、推理模型 Token 截斷防禦、寫入前強校驗防寫閘與併發預算控制。Thu, 22 Jan 2026 00:00:00 GMThttps://haoweichan.github.io/blog/agentic-ingestion-pipeline-part1/https://haoweichan.github.io/blog/agentic-ingestion-pipeline-part1/How an unstructured financial podcast data pipeline evolved from a single prompt into a decoupled multi-agent topology using LangGraph.Thu, 15 Jan 2026 00:00:00 GMThttps://haoweichan.github.io/zh-tw/blog/agentic-ingestion-pipeline-part1/https://haoweichan.github.io/zh-tw/blog/agentic-ingestion-pipeline-part1/從單一 Prompt 瓶頸到 LangGraph 多 Agent 協作：非結構化財經 Podcast 語音與逐字稿 Pipeline 的解耦與拓撲設計。Thu, 15 Jan 2026 00:00:00 GMT